Poor predictions when deploying a custom model on Arabic

[MohieEldinMuhammad]

Following this link instructions https://github.com/JaidedAI/EasyOCR/blob/master/custom_model.md. I have trained a custom model on my own dataset.

Here is the .yml file I used:

network_params:
  hidden_size: 512
  input_channel: 1
  output_channel: 512
  hidden_size: 512

imgH: 64
imgW: 600

lang_list:
         - 'en'
character_list: "0123456789!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~ abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ٠١٢٣٤٥٦٧٨٩«»؟،؛ءآأؤإئااًبةتثجحخدذرزسشصضطظعغفقكلمنهوىيٱٹپچڈڑژکڭگںھۀہۂۃۆۇۈۋیېےۓە"
number: '1234567890١٢٣٤٥٦٧٨٩٠'

The .py file:

import torch
import torch.nn as nn
import torch.nn.init as init
import torchvision
from torchvision import models
from collections import namedtuple
from packaging import version

def init_weights(modules):
    for m in modules:
        if isinstance(m, nn.Conv2d):
            init.xavier_uniform_(m.weight.data)
            if m.bias is not None:
                m.bias.data.zero_()
        elif isinstance(m, nn.BatchNorm2d):
            m.weight.data.fill_(1)
            m.bias.data.zero_()
        elif isinstance(m, nn.Linear):
            m.weight.data.normal_(0, 0.01)
            m.bias.data.zero_()

class vgg16_bn(torch.nn.Module):
    def __init__(self, pretrained=True, freeze=True):
        super(vgg16_bn, self).__init__()
        if version.parse(torchvision.__version__) >= version.parse('0.13'):
            vgg_pretrained_features = models.vgg16_bn(
                weights=models.VGG16_BN_Weights.DEFAULT if pretrained else None
            ).features
        else: #torchvision.__version__ < 0.13
            models.vgg.model_urls['vgg16_bn'] = models.vgg.model_urls['vgg16_bn'].replace('https://', 'http://')
            vgg_pretrained_features = models.vgg16_bn(pretrained=pretrained).features

        self.slice1 = torch.nn.Sequential()
        self.slice2 = torch.nn.Sequential()
        self.slice3 = torch.nn.Sequential()
        self.slice4 = torch.nn.Sequential()
        self.slice5 = torch.nn.Sequential()
        for x in range(12):         # conv2_2
            self.slice1.add_module(str(x), vgg_pretrained_features[x])
        for x in range(12, 19):         # conv3_3
            self.slice2.add_module(str(x), vgg_pretrained_features[x])
        for x in range(19, 29):         # conv4_3
            self.slice3.add_module(str(x), vgg_pretrained_features[x])
        for x in range(29, 39):         # conv5_3
            self.slice4.add_module(str(x), vgg_pretrained_features[x])

        # fc6, fc7 without atrous conv
        self.slice5 = torch.nn.Sequential(
                nn.MaxPool2d(kernel_size=3, stride=1, padding=1),
                nn.Conv2d(512, 1024, kernel_size=3, padding=6, dilation=6),
                nn.Conv2d(1024, 1024, kernel_size=1)
        )

        if not pretrained:
            init_weights(self.slice1.modules())
            init_weights(self.slice2.modules())
            init_weights(self.slice3.modules())
            init_weights(self.slice4.modules())

        init_weights(self.slice5.modules())        # no pretrained model for fc6 and fc7

        if freeze:
            for param in self.slice1.parameters():      # only first conv
                param.requires_grad= False

    def forward(self, X):
        h = self.slice1(X)
        h_relu2_2 = h
        h = self.slice2(h)
        h_relu3_2 = h
        h = self.slice3(h)
        h_relu4_3 = h
        h = self.slice4(h)
        h_relu5_3 = h
        h = self.slice5(h)
        h_fc7 = h
        vgg_outputs = namedtuple("VggOutputs", ['fc7', 'relu5_3', 'relu4_3', 'relu3_2', 'relu2_2'])
        out = vgg_outputs(h_fc7, h_relu5_3, h_relu4_3, h_relu3_2, h_relu2_2)
        return out

class BidirectionalLSTM(nn.Module):

    def __init__(self, input_size, hidden_size, output_size):
        super(BidirectionalLSTM, self).__init__()
        self.rnn = nn.LSTM(input_size, hidden_size, bidirectional=True, batch_first=True)
        self.linear = nn.Linear(hidden_size * 2, output_size)

    def forward(self, input):
        """
        input : visual feature [batch_size x T x input_size]
        output : contextual feature [batch_size x T x output_size]
        """
        try: # multi gpu needs this
            self.rnn.flatten_parameters()
        except: # quantization doesn't work with this 
            pass
        recurrent, _ = self.rnn(input)  # batch_size x T x input_size -> batch_size x T x (2*hidden_size)
        output = self.linear(recurrent)  # batch_size x T x output_size
        return output

class VGG_FeatureExtractor(nn.Module):

    def __init__(self, input_channel, output_channel=512):
        super(VGG_FeatureExtractor, self).__init__()
        self.output_channel = [int(output_channel / 8), int(output_channel / 4),
                               int(output_channel / 2), output_channel]
        self.ConvNet = nn.Sequential(
            nn.Conv2d(input_channel, self.output_channel[0], 3, 1, 1), nn.ReLU(True),
            nn.MaxPool2d(2, 2),
            nn.Conv2d(self.output_channel[0], self.output_channel[1], 3, 1, 1), nn.ReLU(True),
            nn.MaxPool2d(2, 2),
            nn.Conv2d(self.output_channel[1], self.output_channel[2], 3, 1, 1), nn.ReLU(True),
            nn.Conv2d(self.output_channel[2], self.output_channel[2], 3, 1, 1), nn.ReLU(True),
            nn.MaxPool2d((2, 1), (2, 1)),
            nn.Conv2d(self.output_channel[2], self.output_channel[3], 3, 1, 1, bias=False),
            nn.BatchNorm2d(self.output_channel[3]), nn.ReLU(True),
            nn.Conv2d(self.output_channel[3], self.output_channel[3], 3, 1, 1, bias=False),
            nn.BatchNorm2d(self.output_channel[3]), nn.ReLU(True),
            nn.MaxPool2d((2, 1), (2, 1)),
            nn.Conv2d(self.output_channel[3], self.output_channel[3], 2, 1, 0), nn.ReLU(True))

    def forward(self, input):
        return self.ConvNet(input)

class ResNet_FeatureExtractor(nn.Module):
    """ FeatureExtractor of FAN (http://openaccess.thecvf.com/content_ICCV_2017/papers/Cheng_Focusing_Attention_Towards_ICCV_2017_paper.pdf) """

    def __init__(self, input_channel, output_channel=512):
        super(ResNet_FeatureExtractor, self).__init__()
        self.ConvNet = ResNet(input_channel, output_channel, BasicBlock, [1, 2, 5, 3])

    def forward(self, input):
        return self.ConvNet(input)

class BasicBlock(nn.Module):
    expansion = 1

    def __init__(self, inplanes, planes, stride=1, downsample=None):
        super(BasicBlock, self).__init__()
        self.conv1 = self._conv3x3(inplanes, planes)
        self.bn1 = nn.BatchNorm2d(planes)
        self.conv2 = self._conv3x3(planes, planes)
        self.bn2 = nn.BatchNorm2d(planes)
        self.relu = nn.ReLU(inplace=True)
        self.downsample = downsample
        self.stride = stride

    def _conv3x3(self, in_planes, out_planes, stride=1):
        "3x3 convolution with padding"
        return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride,
                         padding=1, bias=False)

    def forward(self, x):
        residual = x

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)

        if self.downsample is not None:
            residual = self.downsample(x)
        out += residual
        out = self.relu(out)

        return out

class ResNet(nn.Module):

    def __init__(self, input_channel, output_channel, block, layers):
        super(ResNet, self).__init__()

        self.output_channel_block = [int(output_channel / 4), int(output_channel / 2), output_channel, output_channel]

        self.inplanes = int(output_channel / 8)
        self.conv0_1 = nn.Conv2d(input_channel, int(output_channel / 16),
                                 kernel_size=3, stride=1, padding=1, bias=False)
        self.bn0_1 = nn.BatchNorm2d(int(output_channel / 16))
        self.conv0_2 = nn.Conv2d(int(output_channel / 16), self.inplanes,
                                 kernel_size=3, stride=1, padding=1, bias=False)
        self.bn0_2 = nn.BatchNorm2d(self.inplanes)
        self.relu = nn.ReLU(inplace=True)

        self.maxpool1 = nn.MaxPool2d(kernel_size=2, stride=2, padding=0)
        self.layer1 = self._make_layer(block, self.output_channel_block[0], layers[0])
        self.conv1 = nn.Conv2d(self.output_channel_block[0], self.output_channel_block[
                               0], kernel_size=3, stride=1, padding=1, bias=False)
        self.bn1 = nn.BatchNorm2d(self.output_channel_block[0])

        self.maxpool2 = nn.MaxPool2d(kernel_size=2, stride=2, padding=0)
        self.layer2 = self._make_layer(block, self.output_channel_block[1], layers[1], stride=1)
        self.conv2 = nn.Conv2d(self.output_channel_block[1], self.output_channel_block[
                               1], kernel_size=3, stride=1, padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(self.output_channel_block[1])

        self.maxpool3 = nn.MaxPool2d(kernel_size=2, stride=(2, 1), padding=(0, 1))
        self.layer3 = self._make_layer(block, self.output_channel_block[2], layers[2], stride=1)
        self.conv3 = nn.Conv2d(self.output_channel_block[2], self.output_channel_block[
                               2], kernel_size=3, stride=1, padding=1, bias=False)
        self.bn3 = nn.BatchNorm2d(self.output_channel_block[2])

        self.layer4 = self._make_layer(block, self.output_channel_block[3], layers[3], stride=1)
        self.conv4_1 = nn.Conv2d(self.output_channel_block[3], self.output_channel_block[
                                 3], kernel_size=2, stride=(2, 1), padding=(0, 1), bias=False)
        self.bn4_1 = nn.BatchNorm2d(self.output_channel_block[3])
        self.conv4_2 = nn.Conv2d(self.output_channel_block[3], self.output_channel_block[
                                 3], kernel_size=2, stride=1, padding=0, bias=False)
        self.bn4_2 = nn.BatchNorm2d(self.output_channel_block[3])

    def _make_layer(self, block, planes, blocks, stride=1):
        downsample = None
        if stride != 1 or self.inplanes != planes * block.expansion:
            downsample = nn.Sequential(
                nn.Conv2d(self.inplanes, planes * block.expansion,
                          kernel_size=1, stride=stride, bias=False),
                nn.BatchNorm2d(planes * block.expansion),
            )

        layers = []
        layers.append(block(self.inplanes, planes, stride, downsample))
        self.inplanes = planes * block.expansion
        for i in range(1, blocks):
            layers.append(block(self.inplanes, planes))

        return nn.Sequential(*layers)

    def forward(self, x):
        x = self.conv0_1(x)
        x = self.bn0_1(x)
        x = self.relu(x)
        x = self.conv0_2(x)
        x = self.bn0_2(x)
        x = self.relu(x)

        x = self.maxpool1(x)
        x = self.layer1(x)
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)

        x = self.maxpool2(x)
        x = self.layer2(x)
        x = self.conv2(x)
        x = self.bn2(x)
        x = self.relu(x)

        x = self.maxpool3(x)
        x = self.layer3(x)
        x = self.conv3(x)
        x = self.bn3(x)
        x = self.relu(x)

        x = self.layer4(x)
        x = self.conv4_1(x)
        x = self.bn4_1(x)
        x = self.relu(x)
        x = self.conv4_2(x)
        x = self.bn4_2(x)
        x = self.relu(x)

        return x

class Model(nn.Module):

    def __init__(self, input_channel, output_channel, hidden_size, num_class):
        super(Model, self).__init__()
        """ FeatureExtraction """
        self.FeatureExtraction = ResNet_FeatureExtractor(input_channel, output_channel)
        self.FeatureExtraction_output = output_channel  # int(imgH/16-1) * 512
        self.AdaptiveAvgPool = nn.AdaptiveAvgPool2d((None, 1))  # Transform final (imgH/16-1) -> 1

        """ Sequence modeling"""
        self.SequenceModeling = nn.Sequential(
            BidirectionalLSTM(self.FeatureExtraction_output, hidden_size, hidden_size),
            BidirectionalLSTM(hidden_size, hidden_size, hidden_size))
        self.SequenceModeling_output = hidden_size

        """ Prediction """
        self.Prediction = nn.Linear(self.SequenceModeling_output, num_class)

    def forward(self, input, text):
        """ Feature extraction stage """
        visual_feature = self.FeatureExtraction(input)
        visual_feature = self.AdaptiveAvgPool(visual_feature.permute(0, 3, 1, 2))  # [b, c, h, w] -> [b, w, c, h]
        visual_feature = visual_feature.squeeze(3)

        """ Sequence modeling stage """
        contextual_feature = self.SequenceModeling(visual_feature)

        """ Prediction stage """
        prediction = self.Prediction(contextual_feature.contiguous())

        return prediction

Then i predict like this:

ar_reader = easyocr.Reader(['en'], recog_network='arabic')
ar_reader.readtext(image_path,paragraph=True)

[amroghoneim]

Same Here!

[ftmasadi]

Even I get very bad results on the training images that I had taken with 100% accuracy Is it the same for you or do you get good accuracy on the training data in the test phase? @MohieEldinMuhammad @amroghoneim

[Arunavameister]

@amroghoneim @ftmasadi @MohieEldinMuhammad Any updates guys? I have a similar problem.

[MohieEldinMuhammad]

Even I get very bad results on the training images that I had taken with 100% accuracy Is it the same for you or do you get good accuracy on the training data in the test phase? @MohieEldinMuhammad @amroghoneim

actually i was overfitting on 2 images only to test the whole pipeline first, so the problem now is that the fine-tuned model is not working properly on deployment phase, hence no need to wast time in training unless the problem is solved

[MohieEldinMuhammad]

@amroghoneim @ftmasadi @MohieEldinMuhammad Any updates guys? I have a similar problem.

No, unfortunately.

[ftmasadi]

@amroghoneim @ftmasadi @MohieEldinMuhammad Any updates guys? I have a similar problem.

May I ask what exactly is your problem? And in which part of your results are there problems? @Arunavameister

[ftmasadi]

Even I get very bad results on the training images that I had taken with 100% accuracy Is it the same for you or do you get good accuracy on the training data in the test phase? @MohieEldinMuhammad @amroghoneim

actually i was overfitting on 2 images only to test the whole pipeline first, so the problem now is that the fine-tuned model is not working properly on deployment phase, hence no need to wast time in training unless the problem is solved

Thank you for your reply But I still don't understand what I should do to solve my problem, even though I spent a lot of time on it. Is it possible for you to explain more about this problem? Thank you very much for your help @MohieEldinMuhammad

[Arunavameister]

@ftmasadi I am getting random results during test predictions even though in training I had about 75% accuracy. I will post an update if I manage to make it work

[MohieEldinMuhammad]

@Arunavameister thanks for the help, i hope you will figure it out 🙏

[ftmasadi]

Thank you very much for your reply. What language do you work on? Because I think this problem is mostly in languages ​​that have a continuous structure ​​like Arabic, Farsi, and Urdu because I didn't encounter such a problem when I used the English database of the site. What is your opinion about this? @Arunavameister

[masoudMZB]

Facing same issue . I am trying to solve this problem. I'll report here if I handled this problem.

[ftmasadi]

thanks for the help, I still don't understand what I should do to solve this problem @masoudMZB

[hayderkharrufa]

Facing same issue. Any updates on this?

[masoudMZB]

@hayderkharrufa @ftmasadi there are many reasons why you can't get proper ouput

• characters ordering in config file
• different backbone
• etc. by the way I suggest you to forget this repo and easyocr check paddleocr

[vajos]

Do you guys have any updates? Im also getting 85% accuracy when training. And then when I try to use the model with THE SAME pictures as I used for training/evaluation I get much lower accuracy.